Issue #22: Meta dangerous Voice AI, leaked Amazon AI doc, AI analytics
Now you see it.
Now you don’t.
Meta recently announced the release of Voicebox, their AI tool for speech generation.
Then decided it was too dangerous and put Voicebox back in the box.
Meta says their text-to-audio tool, Voicebox, can:
Produce audio clips of speech in six languages.
Replicate somebody’s voice using just a 2-second sample audio file.
Regenerate clean audio clips from those disrupted by unwanted noise, like a dog barking.
Complete tasks beyond what it was “specifically trained to accomplish”.
All great benefits for marketers creating audio content.
Being able to turn a short clip of the company CEO or celebrity representative into authentic-sounding personalized recordings could significantly improve your marketing, sales, and customer experience.
So, why the reversal?
There are plenty of “sample” audio clips of the most powerful people in the world freely available online.
And there are plenty of people with questionable motives and ethics to take those audio clips and create unlimited clones for various uses—none which are probably very good.
Do we really need more clips of politicians saying insane things? Not only do they do a good job sounding like insane asylum regulars as it is, we don’t need more of that, please.
Or, imagine a “leaked” audio of a CEO talking about how their product is failing or their financials are in tatters.
You can destroy a company overnight with a fake audio clip.
So for now, Meta is releasing a research paper and a classifier that can identify whether an audio clip is authentic or created by Voicebox, which they say will act as mitigation against misuse of the tool (if and when it’s finally released to the public).
We’ll see how long Voicebox remains locked up.
If you’ve been paying attention to this space at all, you know that someone else will release an equally powerful tool any day now.
And Meta isn’t the only big tech company thinking about AI ethics and regulation:
Google has warned staff about putting confidential information into chatbots, including its own Bard, according to Reuters.
The European Parliament has passed the AI Act (which seems to make life very difficult for those developing AI in Europe—it’s incredibly restrictive on what’s “allowed”).
The UK has announced they have £100 million (~$1.3M) to spend on AI safety, and they’re looking for AI specialists and researchers to help spend it.
Let’s leave the politics behind and discover the latest AI developments that promise to boost your marketing today.
In today’s issue:
Getting started with predictive consumer behavior and deep analytics.
Using AI to analyze customer reviews for marketing campaigns.
Amazon’s leaked document with 67 ways to use ChatGPT and AI.
Alphabet releasing more advertising AI tools (that will take your job).
Google gives clothes shopping a bionic makeover.
Let’s dive in.
Being able to predict consumer behavior is at the top of most marketers’ wish lists.
And there’s an AI for that.
(Actually, there are several ways you can use AI and Machine Learning for this).
Meet Pecan AI, a predictive analytics platform that shows you which customers are worth acquiring more of—and a lot more.
Pecan prepares your raw data for AI models and delivers predictions within days.
There’s also Olvy, a tool that will deliver insights on customer feedback with their AI Copilot for Product Managers (and marketers can use it, too).
And if you’re still waiting on the Code Interpreter mode for ChatGPT+ users, there’s always Rows.com AI. It’s like ChatGPT inside a spreadsheet.
As I’m sure you’ve read in these emails already—but it bears repeating:
AI and Machine Learning is going into everything, everywhere.
And if you’re not a technical marketer or copywriter already, you need to at least become familiar with these tools.
Right now, marketers for your competitors, or other copywriters, are figuring out how to get ChatGPT to write content and copy that requires little or not editing.
They’re using tools like Rows or Pecan to perform analytical tasks for their data (like ad campaigns, emails, content metrics, etc.) that will in turn help them create better performing marketing.
Many of them are not technical. They’re not developers. They don’t “do data”.
And they don’t have to.
These tools are doing much if not all of the work for them.
And they’ll have an advantage over any marketer or copywriter who doesn’t take a few hours in a week to learn to do the same.
There’s an abundance of open-source tools, API scripts, low-code solutions and tools.
And an abundance of free tutorials and how-to’s anywhere you care to look.
You’re facing zero technical barriers to entry.
What’s stopping you?
All you need is a willingness to watch a few tutorials and try the tools.
Perhaps I’ll start creating tutorials, too. If so, what would you like to learn? Hit reply and let me know.
This infographic breaks down Generative AI for you.
It’s a good intro, recap, and filled with interesting statistics.
And if you’re doing paid traffic with Google, you should know that they’re launching more AI-powered features that will, well, do a huge part of your job for you.
One of the new features, called Demand Gen, will use AI to place an advertiser's photo and video ads across several products such as Gmail, the YouTube feed and Shorts.
AI will remove the need for you to think about where you should place your ads, and the technology will focus on finding placements for you.
Google wants to use AI to remove some of the "grunt work" for advertisers—so you can focus more on their marketing strategy and storytelling.
This trend is true for every marketer and copywriter:
A lot of the “grunt work” will be done by AI and Machine Learning.
Same goes for SEO, copywriting, content marketing—basically all of it—is changing, too.
At some point, yes, jobs will either be replaced or thoroughly changed. Nothing will be untouched.
Even a company like Amazon, who is deeply immersed in AI and ML already, are wondering what the heck to do with it all.
A leaked document of Amazon's ideas for using ChatGPT and AI at work lists 67 ways to take advantage of ChatGPT.
Some interesting takeaways, which a lot of marketers and copywriters can identify with:
Amazon employees want to use the AI chatbot to auto-generate software code and marketing materials.
Salespeople could use it to quickly scour financial reports from Amazon and its competitors to find their strategic goals.
It could also help with customer-sentiment analysis, as ChatGPT can "decipher how customers really feel about a product."
There could be a ChatGPT-type search bar for Amazon shoppers that "can explain pros and cons between brands and cite and summarize user reviews.”
Other ideas shared in the impact-and-opportunity analysis document cover everything from auto-generating software code and marketing materials to providing better customer support.
One person wrote that Amazon Comprehend, a natural-language-processing service, feels "very cumbersome" compared with ChatGPT in turning survey data into more meaningful insights. Amazon Lex, a tool that helps build chatbots, can become "too generic" and "boring" compared with other services, another comment said.
And from Andy Jassy, Amazon CEO:
"I would tell you, that in every single one of the businesses and customer experiences in which you work, I believe that you will have transformative generative-AI models as a key part of the customer experience moving forward."
Stop thinking it’s a fad or all hype.
Marketers and copywriters, along with developers, are first to be engulfed in the raging rollout of AI into everything.
Are you at least using ChatGPT every week?
There’s more you can—and should—do.
The smart people over at Bain & Co published a piece on AI for CMOs a few weeks ago. As most articles from big consulting companies, it has its fair share of jargon and corporate-speak.
But it’s worth your time to at least skim it. Because you’ll come across interesting facts and figures like this one:
Source: Bain & Co.
And the money keeps flowing, as you can see below.
$100M: Glean, a high-value startup with a $1 billion valuation, unveils Glean Chat: an enterprise-grade generative AI chat assistant revolutionizing workplace communication.
$28M: Sana, led by a 26-year-old entrepreneur, has launched an AI workplace assistant. Their software integrates eight advanced AI systems, including OpenAI's Whisper, to empower efficient transcription, summarization, and content creation tasks.
$45M: Instabase is looking to transform document processing with AI. The platform aims to automate workflows and content understanding, offering pre-built apps for tasks like income verification and invoice processing.
🤖 Ever feel like you can only get a sensible conversation with yourself?
By analyzing individual user data, Personal AI can create a digital chatbot twin to help you draft messages and memories to share.
With a vast database of 120M parameters, it analyzes incoming information (as chosen by you), and then generates personalized content.
If you’re concerned about that sensitive data getting loose, don’t be: Data collected is only used to train that specific user's model. Ownership of the data remains with the users.
🤖 Jetpack has launched its latest AI-powered assistant.
This WordPress plugin integrates AI directly into the WordPress editor, making content generation quick and effortless.
Jetpack's arsenal includes adaptive writing tones, title and summary generation, content translation, as well as spelling and grammar correction.
Whether you're a blogger, marketer, or business seeking to enhance your content creation process, Jetpack's features deliver engaging results, and WordPress users can enjoy free access.
Otherwise, the plugin offers 20 free requests, and then costs $10 per month for standard CMS accounts.
🤖 Portaly helps you summarize your site with a short and sweet link-in-bio landing page.
With just a single click, you can create a fully customized page compatible with popular platforms like Patreon, Behance, GitHub, Figma, and Notion.
Simply paste your website or profile URL, and let the AI work its magic. It scrapes your site for data, and uses it to generate a personalized page.
With 20 different functional blocks (each fully customizable through an intuitive drag-and-drop interface), you can switch up your own however you see fit.
THE LAST BYTE
Clothes marketing is on the brink of an AI revolution.
And it’s Google who is first out of the closet.
You can now use their virtual try-on (VTO) feature to see what clothes look like on real models with different body shapes and sizes.
The ability to “try before you buy” is a powerful tool in marketing. Especially if it doesn’t involve leaving the comfort of your own home.
The tech is impressive and includes details such as how the fabric drapes, folds, clings, stretches and wrinkles over different shapes and sizes of people.
It uses two images (one of the model and one of the item), takes them apart, and reconfigures them to show the model wearing the item.
It will be interesting to see the sales data on how the products perform for customers using the VTO tool.
The infiltration of retail by AI has unexpected benefits for us as marketers.
Customer reviews are an integral part of shopping on Amazon.
But just think about this from a marketing perspective.
You can use AI to analyze those very same reviews:
Ask it to summarize customer experiences.
Evaluate emotional sentiments, both positive and negative.
Discover how prospects are talking about the problems the product is solving to inspire your sales writing.
Don’t get overwhelmed by the speed of evolution. Get to work.
And if you have specific topics you want me to go in-depth on, let me know.
Last thing before I go, there’s only one image to sum up the news of Mercedes-Benz putting ChatGPT in their cars:
I watched Knight Rider as a kid, all the time.
Good to see KITT becoming a reality.
Anyway, let me reiterate my point from the start:
There are tutorials everywhere on how to create your own tool that produces a full campaign or funnel from a single prompt, how to use API (even if you’re not a developer), how to use AI for analytics, and on it goes.
There are even easy tutorials on how to fine-tune your own LLM for infinite marketing and copy generation, or how to use a vector search for a simple “chatbot” that “talks” to your own marketing materials.
It’s 10x easier to do all of this now than it was even 6 months ago.
I’ve had to do these things the hard way the past 3-4 years. Now I can do it in an afternoon only because there are cheap, easy, and simple tools finally available.
But what’s stopping you from watching these and creating your own stuff?
Why don’t you have a Make.com map already set up that produces a full campaign? Or even just batches of ad or email copy?
Perhaps I’ll start creating tutorials for marketers and copywriters. If so, what would you like to learn?
Hit reply and let me know.
See you next week,